AdMarkup logoAdMarkup
AI Visibility Audit

An AI visibility audit that starts with evidence

An AI visibility audit is a scoped diagnostic that records how AI search systems currently represent a brand, why, and what to change. It combines prompt testing across assistants with crawl access, entity consistency and content structure review.

The problem

Most teams have no idea whether assistants mention them, recommend competitors, or state something false — and conventional analytics will not tell them. Decisions get made on anecdote.

How we approach it

We build a representative prompt set from your real buyer questions, test it across the major assistants, record exactly what is returned and cited, then diagnose the causes and prioritise fixes by expected impact.

Business outcomes

  • A documented baseline of current AI visibility
  • Clear reasons behind absence or misrepresentation
  • A prioritised, costed action plan
  • A repeatable test set for tracking progress

What's included

Prompt set design

We build a representative set from real buyer questions, category comparisons and brand queries, rather than prompts chosen to flatter.

Cross-assistant testing

The set is run across the major assistants and generative search surfaces, with responses, citations and inaccuracies recorded verbatim.

Crawl and retrieval access check

We verify that retrieval crawlers can reach and render your content, and flag bot-management or rendering blocks.

Entity consistency review

Your facts are compared across the site, structured data and major third-party profiles to find contradictions.

Content structure assessment

Priority pages are assessed for extractability: direct answers, self-contained passages, evidence, attribution and freshness.

Prioritised action plan

Findings are ranked by expected impact against effort, with owners and sequencing.

What you receive

  • Prompt set with recorded responses and citations
  • Competitor visibility comparison
  • Crawl and retrieval access findings
  • Entity inconsistency register
  • Content extractability assessment
  • Prioritised action plan
  • Walkthrough session and reusable test set

How we work

  1. 1

    Scope

    Agree buyer questions, competitors and assistants in scope.

  2. 2

    Test

    Run the prompt set and record results verbatim.

  3. 3

    Diagnose

    Establish why competing sources are used instead of yours.

  4. 4

    Prioritise

    Rank fixes by expected impact and effort.

  5. 5

    Hand over

    Present findings and leave the test set for ongoing tracking.

How we measure it

  • Baseline appearance rate per assistant
  • Accuracy of descriptions returned
  • Competitor appearance rate for comparison
  • Recorded factual errors and their likely sources

Honest limitations

Assistant responses vary by user, session and model version, so an audit is a sample rather than a complete measurement. It also has no effect by itself — the value depends on implementing the plan.

Frequently asked questions

How many prompts do you test?
Enough to be representative of your real buying questions rather than a fixed number, typically several dozen across brand, category, comparison and problem-led phrasing.
Can you compare us to competitors?
Yes. The same prompt set is run for named competitors so appearance and description quality can be compared directly.
Will results be the same if we re-run them?
Not exactly, and that is worth understanding upfront. Responses vary between sessions, which is why we log results over time rather than relying on one test.
What if the audit shows we are absent everywhere?
Then the plan tells you why and what it would take, including cases where the honest answer is that conventional SEO and content depth must come first.

Related services

Let's build your growth system.

Tell us where you are now and what you need to reach. We'll come back with scope, sequencing and what we'd do first.